Abstract
The level of variation in hydrogen sulfide formation as a function of yeast genetic background was evaluated for 170 strains, comprised of 169 commercial strains and one native isolate, using six different media: BiGGY agar (assess basal levels of sulfite reductase), synthetic grape juice media with differing nitrogen and vitamin content, and Syrah grape juice. Statistical and cluster analysis placed the strains into 15 clusters, ranging from 1 to 43 members. Sulfide production was not well correlated with the nitrogen content of the medium. In general, sulfide formation was highest in the Syrah juice, in both total number of strains producing the compound and in level of production. There was no correlation between colony color on BiGGY agar and level of production of H2S in any of the liquid media. Strains producing high levels of H2S in Syrah also tended to produce H2S in at least one of the synthetic media. Many moderate to low level producers of H2S in Syrah produced trace to undetectable levels of H2S in synthetic media. Thus, synthetic media can be used to identify strains with the potential to produce high levels of H2S during fermentation. However, a low level of sulfide production in synthetic juices does not indicate low level of production in juice, and strain evaluation in actual juices is still necessary.
The central role of Saccharomyces cerevisiae metabolism in the formation of hydrogen sulfide (H2S) during grape juice fermentation has been well established (Acree et al. 1972, Eschenbruch 1974, Eschenbruch et al. 1978, Giudici and Kunkee 1994, Jiranek et al. 1995a, Linderholm et al. 2008, Rankine 1963, Rauhut and Kurbel 1994, Walker and Simpson 1993). Several factors influence the amount of H2S produced during fermentation and its retention, including levels of elemental sulfur on the grapes at harvest (Rauhut and Kurbel 1994), presence of sulfur dioxide during fermentation (Nickerson 1953, Schutz and Kunkee 1977, Stratford and Rose 1985) or organic compounds containing sulfur (Acree et al. 1972), nitrogen limitation (Giudici and Kunkee 1994, Jiranek et al. 1995a, 1996), and vitamin deficiency (Bohlscheid et al. 2007, Tokuyama et al. 1973, Wainwright 1970, Wang et al. 2003). The timing of formation is also important, as gaseous H2S may be driven from the fermentation via the carbon dioxide stream.
Hydrogen sulfide production levels vary across different yeast genetic backgrounds in response to these conditions (Acree et al. 1972, Mendes-Ferreira et al. 2002, Spiropoulos et al. 2000). This variation has been attributed to the ability to incorporate reduced sulfur into organic compounds and suggests that differences in internal enzyme regulation and activity affect H2S production (Jiranek et al. 1995b, 1996, Linderholm et al. 2006, 2008, Spiropoulos and Bisson 2000). An earlier evaluation of H2S formation across a set of native isolates documented that sulfide production varied over a 100-fold range (Spiropoulos et al. 2000). The aim of this research was to similarly evaluate the differences in H2S formation across a population of commercial strains as compared to one nonproducing native isolate, UCD932.
Nutritional factors are important in the appearance of H2S during fermentation. H2S production has been reported to be elevated in nitrogen-deficient juices (Vos and Gray 1979, Giudici and Kunkee 1994, Jiranek et al. 1995a, Wang et al. 2003). However, some native isolate strains of Saccharomyces have produced elevated levels of H2S as a response to high levels of nitrogen as well as low, depending on the strain (Mendes-Ferreira et al. 2009, Spiropoulos et al. 2000). It has been demonstrated that the time at which nitrogen is depleted is significant and that the highest levels of H2S were produced when nitrogen supply was exhausted during the rapid growth phase (Jiranek et al. 1995a). Commercial strains have been selected for traits that are most compatible with fermentation, and although there has not necessarily been a deliberate selection for reduction of sulfide formation in response to nitrogen, this trait may be a feature of commercial strains. The goal of this study was to determine if the previously observed variation seen in native isolates (Spiropoulos et al. 2000) is similar to the variation across commercial strains.
The total or free amino nitrogen concentration affects H2S production, and the balance or type of nitrogen sources is also a significant factor. Available ammonia nitrogen, alone, is not adequate to control for H2S production. Addition of most amino acids such as arginine, lysine, aspartate, serine, and leucine suppressed H2S formation, while the addition of cysteine or threonine increased H2S production (Jiranek et al. 1995a, Rauhut 1992).
Deficiencies in vitamins and micronutrients essential for the synthesis of sulfur-containing amino acids may also contribute to H2S production. In one study, strains of Saccharomyces produced increased H2S in media deficient in pantothenate and vitamin B6 (pyridoxine) (Wainwright 1970). Pantothenate is a component of coenzyme A, which is required in the synthesis of O-acetylhomoserine from homoserine. Vitamin B6 is also required for the biosynthesis of cysteine and methionine (Wainwright 1970). Deficiencies in vitamin B6, and the accompanying failure to synthesize adequate amounts of sulfur-containing amino acids, may lead to induction of the pathway, as these compounds repress sulfate reduction (Mountain et al. 1991).
Various environmental and fermentation conditions also influence the amount of hydrogen sulfide that is produced or retained. Fermentation temperature (Rankine 1963, Karagiannis and Lanaridis 1999), juice turbidity (Karagiannis and Lanaridis 1999), and level of soluble solids and titratable acidity (Vos and Gray 1979) have been shown to significantly affect H2S levels. The presence of various constituents such as metal ions have also been suggested to result in increased H2S in wine, although this has not been conclusively demonstrated (Eschenbruch 1974). High-level additions of the antimicrobial compound sulfite (SO2) have also been reported to result in increased formation of H2S (Karagiannis and Lanaridis 1999). Some of these effects may be indirect: that is, affect the amount of H2S remaining in the wine in the volatile form.
The goal of this study was to evaluate the impact of nitrogen level and vitamin limitation on H2S formation across a genetically diverse population of commercial wine strains of S. cerevisiae for comparison to an earlier similar study of native isolates (Spiropoulos et al. 2000). Statistical and cluster analyses were performed in order to identify groupings of strains with similar metabolic behavior. The specific aim of this research was to identify fermentation conditions that could be used to predict formation of H2S during grape juice fermentation.
Materials and Methods
Yeast strains.
All 170 isolates of S. cerevisiae were obtained from the Wine Microbe Collection of the University of California Davis (UCD) Department of Viticulture and Enology, 50 of which were provided by Lallemand (Lallemand, Montreal, QE). The strains used were UCD715, UCD750, UCD756, UCD763, UCD764, UCD770, UCD772, UCD777, UCD778, UCD779, UCD781, UCD804, UCD810, UCD813, UCD818, UCD829, UCD866, UCD867, UCD869, UCD904, UCD905, UCD907, UCD908, UCD909, UCD924, UCD925, UCD926, UCD927, UCD928, UCD929, UD932, UCD959, UCD960, UCD961, UCD963, UCD964, UCD965, UCD966, UCD967, UCD968, UCD969, UCD972, UCD974, UCD975, UCD976, UCD977, UCD978, UCD979, UCD980, UCD1047, UCD2032, UCD2033, UCD2034, UCD2035, UCD2036, UCD2038, UCD2039, UCD2061, UCD2068, UCD2069, UCD2070, UCD2071, UCD2072, UCD2073, UCD2074, UCD2099, UCD2212, UCD2214, UCD2216, UCD2389, UCD2390, UCD2391, UCD2392, UCD2393, UCD2394, UCD2395, UCD2410, UCD2411, UCD2412, UCD2413, UCD2414, UCD2415, UCD2416, UCD2417, UCD2496, UCD2497, UCD2498, UCD2499, UCD2500, UCD2501, UCD2502, UCD2521, UCD2522, UCD2523, UCD2524, UCD2525, UCD2526, UCD2527, UCD2528, UCD2529, UCD2530, UCD2531, UCD2532, UCD2533, UCD2534, UCD2535, UCD2536, UCD2537, UCD2538, UCD2539, UCD2540, UCD2541, UCD2542, UCD2543, UCD2544, UCD2545, UCD2546, UCD2553, UCD2554, and UCD2555. Strains contributed by Lallemand (with Lallemand strain designations in parenthesis) were: UCD2556 (A145), UCD2557 (B248), UCD2558 (C85), UCD2559 (D89), UCD2560 (E119), UCD2561 (F103), UCD2562 (G31), UCD2563 (H72), UCD2564 (I74), UCD2565 (J73), UCD2566 (K93), UCD2567 (L90), UCD2568 (M82), UCD2569 (N51), UCD2570 (O54), UCD2571 (P51), UCD2572 (Q53), UCD2573 (R39), UCD2574 (S59), UCD2575 (T22), UCD2576 (U83), UCD2577 (V44), UCD2578 (W94), UCD2579 (X22), UCD2580 (Y68), UCD2581 (Z46), UCD2582 (AA89), UCD2583 (BB111), UCD2584 (CC96), UCD2585 (DD86), UCD2586 (EE46), UCD2587 (FF114), UCD2588 (GG89), UCD2589 (HH59), UCD2590 (II47), UCD2591 (JJ28), UCD2592 (KK29), UCD2593 (LL45), UCD2594 (MM43), UCD2595 (NN37), UCD2596 (OO54), UCD2597 (PP18), UCD2598 (QQ15), UCD2599 (RR8), UCD2600 (SS94), UCD2601 (TT49), UCD2602 (UU9), UCD2603 (VV14), UCD2604 (WW43), and UCD2605 (XX53).
Media.
Synthetic minimal must media (Triple M) and Syrah grape juice from the 2007 UCD harvest were the two types of media used. Triple M was prepared as described elsewhere (Spiropoulos et al. 2000) and modified to achieve three nitrogen concentrations of 433, 123, and 55 N mg/L and a reduced vitamin concentration at 55 N mg/L. Details of media components changed to achieve the above concentrations are shown (Table 1⇓). The N mg/L value was calculated as before for yeast usable nitrogen (Spiropoulos et al. 2000) and is based on direct measurement of the amino acid content of the completed synthetic medium by the UCD Molecular Structure Facility. Proline was not included in the N mg/L value, as it cannot be utilized as a nitrogen source under anaerobic conditions. Histidine and lysine likewise are not utilized as nitrogen sources by S. cerevisiae, so they were counted as contributing a single N equivalent since they can be used directly as an amino acid (Monteiro and Bisson 1991). Arginine degradation yields one molecule of praline, so only three of the four N molecules in arginine are available for use by Saccharomyces. Therefore, only three N equivalents for arginine were used in the summation.
Triple M and the Syrah grape juice were filter-sterilized with a 0.45-μm filter. The grape juice was used without any further modifications. The Syrah juice was used in control fermentations to test for spontaneous fermentation (failure of sterile filtration or recontamination) and for release of H2S due to nonmicrobial processes. There were no contaminating organisms in the Syrah, indicating the sterile filtration was successful. Likewise, no H2S developed in the uninoculated control fermentations, indicating that the juice did not contain chemical precursors of H2S capable of yielding a dark area on the detection column in the absence of yeast activity. The nitrogen concentration of Syrah grape juice was analyzed by the UCD Molecular Structure Facility via high-pressure liquid chromatography and was 408 N mg/L (Table 2⇓). BiGGY agar was obtained from Becton, Dickinson, and Company (Sparks, MD) and prepared according to manufacturer’s instructions.
H2S screen.
Starter cultures of all strains were prepared in 5 mL of appropriate media and incubated at room temperature with agitation of 100 rpm on a roller drum for 36 to 48 hr. H2S detection tubes were prepared by inserting rolled 3 mm Whatman filter paper into a disposable plastic pipette. The paper was saturated with 250 μL of 3% lead acetate solution, and these tubes were inserted into silicon stoppers. The strip detection technology was optimized in preliminary experiments. Volume and percentage of solution were varied for both high and low producing strains and reproducibility evaluated over 10 or more simultaneously run fermentations and compared to those of the same strain run using different batches of media. The variations in level of darkening were slight largely because of the continual mixing achieved on the roller drum. This process leads to a more uniform and complete release of sulfide from the fermentations and yields far less variability than static cultures.
Fermentations were conducted in 10 mL media in 16 mm × 150 mm culture tubes. The media was inoculated with the appropriate strains to an OD650 of 0.03 and incubated for 72 hr with the same conditions as the starter cultures. Relative levels of H2S were determined by measuring the brown precipitate in the detection tube in millimeters. Duplicate or triplicate fermentations were run, depending upon the strain, the conditions, and the reproducibility of media, which was generally well within 20%. The estimate of variance was reported as the range of values around the mean rather than the standard deviation, because the standard deviation, as expected, was less than the range value. Since this is a semiquantitative method, the data was evaluated using the higher degree of variance.
Statistical analyses.
Statistical analysis was performed by correspondence analysis and clustering using SAS (SAS Institute Inc., Cary, NC). Correspondence analysis is a computational method for the study of association between variables (Greenacre 1984, 2007). It is similar to principal component analysis, but for qualitative data, and it displays a low dimension projection of the data on a plane (the biplot). The aim is to embed both rows (strains) and columns (different media conditions) of a matrix in the same space, thus revealing associations between them. In correspondence analysis, the points are the sum of the distances of the points to their centroid, which is proportional to the value of the X2 statistics of the data table. Similar objects are therefore clustered together. The farther away from the centroid (the intersection of the lines), the more pronounced the association of these strains with this condition. However, the distance between a strain and condition cannot be directly interpreted.
Cluster analysis is a tool to reveal groups or subsets that are similar in some sense without explaining why they exist. Hierarchical clustering using the agglomerative average linkage method was used. Cluster to cluster distance is defined here as the average distance between all members of one cluster and all members of another cluster.
Results
Effect of N concentration on H2S formation.
The high (433) N mg/L used in this study was identical to the medium used in the earlier study to evaluate H2S production across a set of native isolates (Spiropoulos et al. 2000). Analysis of juice nitrogen requirements (Monteiro and Bisson 1991) indicated that a value of 120 N mg/L calculated in the same way was sufficient for most juices and growth conditions to ensure completion of the fermentation but did not support as rapid of a fermentation as a higher N content. This value was therefore chosen as a sufficient yet low level of nitrogen, one that is not deficient for fermentation progression but that may induce some physiological stress. The very low 55 N mg/L was chosen to impose nitrogen limitation. In preliminary studies, reducing the micronutrient content of the 433 and 123 N mg/L synthetic juices did not impact sulfide formation or fermentation rate. Reduction of the micronutrient levels at 55 N mg/L did impact fermentation rate and imposed a further stress on the cultures, but most commercial strains were still able to complete fermentation at a slower rate.
The pattern of formation of H2S by 170 strains in five liquid media and on BiGGY indicator agar indicated a variety of responses to the different media (Supplemental Table 1). Of the 170 strains, seven did not produce significant levels of H2S on any of the media evaluated: UCD867, UCD908, UCD909, UCD2500, UCD2522, UCD2584, and the comparison native isolate strain UCD932. These strains showed a variety of colony colors on BiGGY agar, ranging from white and light tan to brown. A total of 163 strains (96%) produced detectable H2S in the Syrah grape juice. This juice was not deficient in nitrogen, with a nitrogen level near that of the high nitrogen synthetic juice (Table 2⇑). In contrast, only 110 strains produced H2S in the synthetic juice medium that had a nitrogen level comparable to that of the Syrah juice. The cluster analysis of the sulfide production data indicated there was no significant difference between undetected production and trace amounts less than 1.0 mm. By this criterion, 124 (73%) of the strains evaluated were hydrogen sulfide formers in the Syrah juice and only 31 (18%) in the high nitrogen synthetic juice (Table 3⇓). The synthetic media with the highest number of sulfide producers, 33 strains (19% of all strains), was the middle nitrogen level of 123 N mg/L. Of the 31 strains with sulfide formation in the presence of high nitrogen concentration (433 N mg/L), 19 also produced H2S at 123 N mg/L. Only 12 of the 170 strains showed the pattern of reduction of H2S formation with increasing nitrogen (comparing 123 N mg/L to 433 N mg/L): UCD960, UCD961, UCD978, UCD2033, UCD2036, UCD2214, UCD2502, UCD2525, UCD2544, UCD2573, UCD2593, and UCD2598. Thus, less than 7% of all strains tested would be expected to respond to nitrogen addition with a decrease in sulfide formation. In addition, the finding that most strains will produce H2S in filtered-sterilized Syrah juice suggests that juice factors other than nitrogen content are more important.
Some of the strains evaluated were different accessions of the same commercial strain. These accessions were from different years and lots of production, so it is not clear that the strains would be identical in H2S production behavior. Three strains were classified as Epernay II (2): UCD750, UCD813, and UCD929. All three strains had brown to dark brown colony colors on BiGGY agar. Two of the strains were low to non-sulfide producing, UCD813 and UCD929. UCD750 had a darker colony color on BiGGY and produced H2S in the high nitrogen synthetic medium and in Syrah juice (Supplemental Table 1). The 1118 strains (UCD756, UCD770, UCD829) showed only trace levels of H2S in the synthetic media, but tended to produce sulfide in the Syrah juice. The Montrachet strains (UCD810, UCD928) showed nearly identical patterns of sulfide formation that were similar to those previously determined for UCD522 (Spiropoulos et al. 2000).
Effect of micronutrient limitation on H2S formation.
Deficiencies for micronutrients have also been reported to result in increased H2S formation under fermentative conditions (Bohlscheid et al. 2007, Tokuyama et al. 1973, Wainwright 1970). The impact of reduction of micronutrients was evaluated in synthetic juice media (Supplemental Table 1). In preliminary tests reduction of micronutrients at higher levels of nitrogen did not impact the levels of sulfide formed. Therefore, the impact of micronutrient limitation was evaluated using a limiting nitrogen concentration, 55 N mg/L. Contrary to what was predicted from the literature, decreasing the micronutrient content generally led to a reduction in sulfide formation (Table 3⇑). Six strains (UCD810, UCD928, UCD2068, UCD2392, UCD2558, and UCD2559) showed an increase in H2S production under micronutrient limitation when compared to the same strain in low nitrogen with sufficient micronutrients, but these strains also produced hydrogen sulfide at other nitrogen levels in synthetic juice and in Syrah. One explanation for the appearance of such a high level of formation of H2S in Syrah is that the juice may have been deficient in micronutrients because it was sterile-filtered. However, the synthetic juice comparisons suggest that micronutrient deficiency is not a factor driving sulfide formation in these strains. However, the rates of fermentation were not significantly slower than the 55 N mg/L medium with full micronutrient supplementation for most strains.
Correlation of colony color on BiGGY agar with H2S formation.
BiGGY agar contains bismuth, which forms a precipitate with sulfide yielding pigmented colonies. The level of brownness of the colony is correlated with the basal level of sulfite reductase activity (Jiranek et al. 1995b). Thus, brown colonies have high levels of sulfite reductase compared with lighter colored colonies. Colony color on BiGGY ranged from white to dark brown (Table 4⇓). The level of H2S formation was evaluated in Syrah juice as a function of colony color on BiGGY. The majority of the strains (118) produced tan to brown colonies. There was no correlation between colony color on BIGGY and production of H2S in Syrah or any of the synthetic juices. The non-H2S producers were evenly distributed among the white to brown colony colors. None of the nonproducers had dark brown colonies on BiGGY agar, suggesting that very high levels of sulfite reductase may indicate a tendency to form sulfide under fermentative conditions. The majority of the very high H2S producers in Syrah displayed tan to brown colonies on BiGGY agar. There appears to be no correlation between colony color on BiGGY agar and tendency to produce H2S in juice or in media.
Statistical and cluster analysis of strain differences.
Initially, the data suggest that all possible patterns of H2S formation were observed across the five liquid media and BiGGY agar. A cluster analysis was performed to determine if there were any natural groupings of the strains (Figure 1⇓). This was followed by taking the strains in each group and evaluating their specific pattern of sulfide formation (Table 5⇓). The correspondence analysis biplot summarizes the 170 strains and five different media conditions in two dimensions and groups strains based on the cluster analysis (Figure 1⇓). The total inertia is 0.64, so all of the dimensions explain 64% of the variance in the data, and of that the percentage of variance explained by the first dimension is 44% and the second dimension is 20%.
The biplot indicates 15 distinct clusters, ranging from a high of 43 members (cluster 8) to a low of one member (clusters 14 and 15) (Figure 1⇑). Several interesting observations can be made from this analysis. First, the seven strains that did not produce sulfide under any liquid growth condition are not segregated as a separate cluster but are grouped with strains showing trace amounts in most media. This finding is consistent with our expectations that a faint gray line on the sulfide detection tubes is not distinguishable from no production or undetectable production. Second, some of the classes that show similar patterns of production are grouped closely on the biplot. For example, clusters 1 and 2, in the lower left quadrant, show some overlap and are somewhat distinguished by the amount of darkening of the H2S detection tube for the Syrah fermentations, but in general cluster 1 strains show a low level (above trace) production of sulfide in other media while cluster 2 strains do not (Table 5⇑).
The media are also displayed on the biplot. BiGGY agar is isolated in the left quadrant (Figure 1⇑). The synthetic juice media cluster on the right, with the micronutrient deficient medium showing the greatest distance from the other three. The two high nitrogen media, 433 N mg/L and Syrah are not close, suggesting that nitrogen content per se is not a driver of the differences in strain behavior observed. The clusters represented by only one or two individuals tend to show high levels of production in some medium other than Syrah.
Discussion
There were two goals in the analysis of H2S of 170 commercial and native isolates: first, to assess the level of variability in H2S across a population of commercial strains in response to nutrient limitation and, second, to determine if nitrogen or micronutrient limitation could be used in synthetic media for the reliable prediction of sulfide production behavior in juice. In comparing this study to a previous study (Spiropoulos et al. 2000), the commercial strains showed similar variability in production of H2S. The filtered red juice used in the earlier study contained moderate nitrogen (264 N mg/L), in which H2S production ranged from undetectable to 14 cm of darkening of the lead acetate column. In the current study, H2S production in the Syrah juice ranged from undetectable to 15 cm of darkening. In the Triple M juice at 433 N mg/L, production of H2S by the native isolates ranged from undetectable to 1 cm of darkening. In contrast the commercial strains showed a range of darkening from undetectable to 9 cm of darkening (UCD927). However, in general the commercial strains showed similar variability to the native isolates in H2S production. They also displayed the same lack of correlation of juice H2S formation and BiGGY colony color. Variability in the allele sequences for genes of the sulfate reduction pathway has been reported (Linderholm et al. 2006), and there is likely significant variation in the activity level of the enzymes of this pathway in the population of commercial strains. Manipulation of the levels of activity of enzymes in this pathway has been associated with alterations of the level of H2S produced (Omura and Shibano 1995, Spiropoulos and Bisson 2000, Swiegers and Pretorius 2007).
With respect to the second goal of evaluating the impact of nutrient limitation, the sterile-filtered red juice yielded the highest number of sulfide producers and the highest levels of sulfide formed across the media conditions. This juice was not nitrogen deficient and fermentations were rapid, suggesting other nutrients were likewise not limiting. None of the media, including the high nitrogen medium that yielded similar fermentation rates, matched the levels of H2S production in the Syrah juice. Other factors, such as stress imposed by phenolic compounds, may explain the higher yields of H2S in the Syrah juice. Again, this is identical to what was previously observed for the native isolates (Spiropoulos et al. 2000) and is consistent with other surveys of H2S production (Mendes-Ferreira et al. 2009). Unfortunately, none of the synthetic media evaluated are of use in predicting strain behavior under fermentation production conditions, and commercial strains will need to continue to be screened in actual juice media.
This analysis did not substantiate the belief that H2S formation is largely driven by nitrogen deficiency. Only a few strains of the 170 showed a decrease in H2S formation with increasing nitrogen content of the medium. What then explains the observations of a decrease in H2S formation with nitrogen addition? It is possible that the observed nitrogen effects are indirect. For example, the faster fermentation rate enabled by sufficient nitrogen may allow the cells to adapt to stress more effectively, thus limiting stress-induced sulfide formation. In addition, the higher nutritional content results in greater carbon availability for energy and adenosine triphosphate production, which also may reduce cellular metabolic stress.
Conclusions
Only 6 of 169 commercial strains screened showed no H2S production in Syrah juice or any other media, similar to native isolate UCD932 which was used as a control. This finding suggests that sulfide formation is common among wine and native isolates of Saccharomyces. Cluster analysis grouped the 170 strains into 15 distinct clusters based on H2S production patterns across synthetic juices of differing nutrient content and in natural grape juice. The synthetic juices evaluated can be used to identify high H2S-producing strains, as strains producing H2S in the synthetic media always produced H2S in the Syrah juice. However, the converse—that strains failing to produce H2S in the synthetic media will fail to produce H2S in actual juices—was not the case.
Footnotes
↵4 (current address) Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720.
Acknowledgments: This research was supported by grants from the American Vineyard Foundation, California Competitive Grants Program for Research in Viticulture and Enology, and the Maynard A. Amerine Endowment.
The authors thank Angela Linderholm and Carrie Findleton for helpful discussions.
- Received August 2009.
- Revision received March 2010.
- Accepted March 2010.
- Published online September 2010
- Copyright © 2010 by the American Society for Enology and Viticulture